Startup profile · funding coverage
First Person Labs funding, valuation and investors
Makes Second, a portable, user-owned personal context model for AI assistants, stored locally by default and shared using permissioned APIs/MCP.
Verified company research · funding, investors and latest developments
Keep track of First Person Labs
Save this profile to your VCT watchlist for a quick return.
Funding, valuation & investors
Answer-first snapshotLatest funding
Seed
$10M · November 17, 2026
Latest known valuation
Undisclosed
Seed · November 17, 2026
Total disclosed equity funding
$10M
Excludes debt, grants, acquisitions, secondaries, and IPO proceeds.
Current status
private
seed · United States (precise headquarters location not independently established)
Investors in latest funding
Lead: Andreessen Horowitz
Other: Mozilla Ventures , Jeff Dean , Ram Shriram , Jeetu Patel , Percy Liang , J. Alex Halderman , Sanmi Koyejo
Overview
First Person Labs PBC, founded in 2026 by Ken Liu and Erik Chi, develops Second: a Mac-first personal user model using background on-device activity to create an inspectable portable context folder for cross-agent MCP and API access. Ken Liu publicly announced a $10M Seed led by Andreessen Horowitz on Nov 17, 2026, with Mozilla Ventures and angels including Jeff Dean, Ram Shriram, Jeetu Patel, Percy Liang, J.
Alex Halderman and Sanmi Koyejo. The original a16z memo by Joel de la Garza and Martin Casado confirms the investment, not the round amount. Second is early-access only; pricing, paying users, revenue, security audit, and independent performance benchmarks are undisclosed. Company describes optional private cloud and inference modes; 'on-device by default' must not be misinterpreted as data never leaving device.
Why First Person Labs is interesting
Co-founder Ken Liu announced a $10M a16z-led Seed on Nov 17 2026, republished by a16z's official LinkedIn account. a16z's investment thesis supports company backing but omits terms. Second is in early access, with no independently validated privacy or commercial evidence.
Product & use cases
Illustrative prospective Mac user journey from founder documentation, not hands-on testing: (1) join early access/waitlist and install Second when invited; (2) review and grant scoped OS activity permissions; (3) allow background agents to summarize recent activities, projects, people and writing into local, user-owned context; (4) inspect, correct, export or back up the local folder; (5) connect selected AI agents through MCP or local/cloud APIs, allowing access per application; (6) evaluate revocation and deletion, distinguish local inference from optional private-cloud sync and hosted inference. Actual commercial availability, full permission UX and security posture not independently validated.
- Personal assistant retrieves up-to-date projects and preferences without repetitive manual instructions.
- Developer connects multiple tools to one user-controlled context layer rather than siloed assistant memories.
- User reviews the saved personal context and scopes what individual agents can read, subject to real enforcement and testing.
Key facts
- $10M Seed led by a16z, disclosed Nov 17 by founder Ken Liu and reposted by official a16z.
- Second's user context lives in a locally inspectable, editable, movable folder, exposed via permissioned local/cloud APIs and MCP.
- Bring a Claude/ChatGPT subscription, personal API key or local inference; optional paid private inference, no published price.
- Early access/waitlist, no audited personal-data security controls, paying-customer metrics or model benchmarks.
Funding history (newest first)
Seed
2026-11-17 $10M Valuation: Undisclosed- Andreessen Horowitz (lead)
- Mozilla Ventures (participant)
- Jeff Dean (participant)
- Ram Shriram (participant)
- Jeetu Patel (participant)
- Percy Liang (participant)
- J. Alex Halderman (participant)
- Sanmi Koyejo (participant)
Source: Read source · x.com
Investors in our directory
Funds linked from First Person Labs's profile — open a fund page for stage focus and related deal articles.
Competitive landscape
Edge: Proposed portable user-owned cross-agent context, versus separate memory inside individual assistants. Security, cost, reliability and data minimization remain unproven.
Continuous OS activity collection creates sensitive data risks involving passwords, medical information, private messaging and bystander content; local folders may be exfiltrated. User permission granularity, auditability, data deletion, optional cloud replication, third-party inference and cross-tool prompt-injection defenses need external testing. Company intends to open-source components, not confirmed delivered. No public audited security review, benchmark, subscription tariff or customer revenue.
-
ChatGPT Memory alternative
Personalization integrated inside one assistant; compare portability and sharing controls.
-
Claude memory alternative
Assistant-integrated project context and personalization.
-
Mem0 direct
Persistent application-agent memory layer for developers.
-
Letta direct
Developer agent-state and memory tooling.
-
Obsidian plus MCP alternative
Manually maintained local file context with explicit operational responsibility.
Market / IPO context
Editorial / static context — not a live quote.
Last updated:
Editorial note: AI tools assisted with research, structure, or drafting. Venture Capital Tracker retains human editorial responsibility for factual accuracy, relevance, and source quality before publication.